Thermoelectric generators have applications in basically any space where you're generating excess waste heat (and you're able to maintain a temperature gradient). Off the top of my head, industrial scale distillation, desalination, garbage incineration, high energy lasers and pyro/electrometallurgy. The advancement of the technology for the automotive space would really have served as a springboard to increase efficiency in other areas.
> thermoelectric generators are simply not efficient enough to adequately offset the heat energy losses in ICEs.
Any amount of waste heat that can be converted to electricity and used or stored is good. Why not try to get every last joule out of your fuel?
You're right that the technology has been inefficient. For typical ceramic thermoelectric generators the conversion ratio from heat to electricity is low, it only works if you can maintain a temperature gradient and you have to contend with self-heating, but addressing the inefficiencies is exactly what JPL would have been trying to do. You couldn't ask for a better partner to push the technology further.